US 7003757 B2
Techniques for fabricating a device include forming a fabrication layout, such as a mask layout, for a physical design layer, such as a design for an integrated circuit, and identifying evaluation points on an edge of a polygon corresponding to the design layer for correcting proximity effects. Techniques include selecting from among all edges of all polygons in a proposed layout a subset of edges for which proximity corrections are desirable. The subset of edges includes less than all the edges. Evaluation points are established only for the subset of edges. Corrections are determined for at least portions of the subset of edges based on an analysis performed at the evaluation points. Other techniques include establishing a projection point on a first edge corresponding to the design layout based on whether a vertex of a second edge is within a halo distance. An evaluation point is determined for the first edge based on the projection point and characteristics of the first edge. It is then determined how to correct at least a portion of the edge for proximity effects based on an analysis at the evaluation point.
1. A method of correcting for proximity effects in a layout, the layout including a first layout for a trim mask and a second layout for a phase-shifting mask, the trim mask and the phase-shifting mask used to fabricate a wafer, the method comprising:
identifying edges of the first and second layouts that will print on the wafer;
evaluating the printing edges of the first and second layouts;
dissecting the printing edges of the first and second layouts to form a plurality of segments;
evaluating only a subset of the plurality of segments;
placing evaluation points on the subset of the plurality of segments; and
correcting only the subset of the plurality of segments.
2. The method of
identifying a first edge of the first and second layouts that will partially print on the wafer;
adding at least one dissection point to the first edge to divide the first edge into a plurality of segments;
placing a first evaluation point on a first segment of the plurality of segments that will print; and
correcting the first segment for proximity effects using the first evaluation point.
3. A method of correcting for proximity effects in a layout, the layout including a first layout for a trim mask and a second layout for a phase-shifting mask, the trim mask and the phase-shifting mask used to fabricate a wafer, the method comprising:
using dissection parameters to select dissection points on an edge of a polygon;
placing an evaluation point between each pair of successive dissection points on the edge; and
determining whether the edge is a last edge of the polygon,
wherein if the edge of the polygon is not a last edge, then determining whether all of the edge is printed on the wafer,
wherein if all the edge is printed, then analyzing the next edge on the polygon.
4. The method of
adding at least one additional dissection point to the edge, thereby forming a plurality of segments; and
moving the evaluation point to an important segment of the plurality of segments.
5. The method of
6. The method of
7. The method of
8. The method of
This application is a divisional of U.S. patent application Ser. No. 09/675,197, entitled “Dissection Of Edges With Projection Points In A Fabrication Layout For Correcting Proximity Effects” filed Sep. 29, 2000 U.S. Pat. No. 6,792,590.
This application is related to U.S. patent application Ser. No. 09/675,582, entitled “Dissection Of Corners In A Fabrication Layout For Correcting Proximity Effects,” filed on Sep. 29, 2000, invented by Christophe Pierrat and Youping Zhang.
This application is related to U.S. patent application Ser. No. 09/676,375, entitled “Dissection Of Printed Edges From A Fabrication Layout For Correcting Proximity Effects,” filed on Sep. 29, 2000, invented by Christophe Pierrat and Youping Zhang.
This application is related to U.S. patent application Ser. No. 09/676,356, entitled “Selection Of Evaluation Point Locations Based on Proximity Effects Model Amplitudes For Correcting Proximity Effects In A Fabrication Layout,” filed on Sep. 29, 2000, invented by Christophe Pierrat and Youping Zhang.
This application is related to U.S. patent application Ser. No. 09/728,885, entitled “Displacing Edge Segments On A Fabrication Layout Based on Proximity Effects Model Amplitudes For Correcting Proximity Effects,” filed on Dec. 1, 2000, invented by Christophe Pierrat and Youping Zhang.
This application is related to U.S. patent application Ser. No. 09/130,996, entitled “Visual Inspection and Verification System,” filed on Aug. 7, 1998.
This application is related to U.S. patent application Ser. No. 09/153,783, entitled “Design Rule Checking System and Method,” filed on Sep. 16, 1998.
This application is related to U.S. patent application Ser. No. 09/544,798, entitled “Method and Apparatus for a Network Based Mask Defect Printability Analysis System,” filed on Apr. 7, 2000.
This application is related to U.S. patent application Ser. No. 09/154,415, entitled “Data Hierarchy Layout Correction and Verification Method and Apparatus,” filed on Sep. 16, 1998.
This application is related to U.S. patent application Ser. No. 09/154,397, entitled “Method and Apparatus for Data Hierarchy Maintenance in a System for Mask Description,” filed on Sep. 16, 1998.
This application is related to U.S. patent application Ser. No. 09/632,080, entitled “General Purpose Shape-Based Layout Processing Scheme for IC Layout Modifications,” filed on Aug. 2, 2000.
1. The Field of the Invention
This invention relates to the field of printed feature manufacturing, such as integrated circuit manufacturing. In particular, this invention relates to automatically identifying evaluation points where errors are computed and analyzed to achieve improved agreement between a design layout and an actual printed feature.
2. Description of Related Art
To fabricate an integrated circuit (IC), engineers first use a logical electronic design automation (EDA) tool, also called a functional EDA tool, to create a schematic design, such as a schematic circuit design consisting of symbols representing individual devices coupled together to perform a certain function or set of functions. Such tools are available from CADENCE DESIGN SYSTEMS and from SYNOPSYS. The schematic design must be translated into a representation of the actual physical arrangement of materials upon completion, called a design layout. The design layout uses a physical EDA tool, such as those available from CADENCE and AVANT!. If materials must be arranged in multiple layers, as is typical for an IC, the design layout includes several design layers.
After the arrangement of materials by layer is designed, a fabrication process is used to actually form material on each layer. That process includes a photo-lithographic process using a mask having opaque and transparent regions that causes light to fall on photosensitive material in a desired pattern. After light is shined through the mask onto the photosensitive material, the light-sensitive material is subjected to a developing process to remove those portions exposed to light (or, alternatively, remove those portions not exposed to light). Etching, deposition, diffusion, or some other material altering process is then performed on the patterned layer until a particular material is formed with the desired pattern in the particular layer. The result of the process is some arrangement of material in each of one or more layers, here called printed features layers.
Because of the characteristics of light in photolithographic equipment, and because of the properties of the material altering processes employed, the pattern of transparent and opaque areas on the mask is not the same as the pattern of materials on the printed layer. A mask design process is used, therefore, after the physical EDA process and before the fabrication process, to generate one or more mask layouts that differ from the design layers. When formed into one or more masks and used in a set of photolithographic processes and material altering processes, these mask layouts produce a printed features layer as close as possible to the design layer.
The particular size of a feature that a design calls for is the feature's critical dimension. The resolution for the fabrication process corresponds to the minimum sized feature that the photolithographic process and the material processes can repeatably form on a substrate, such as a silicon wafer. As the critical dimensions of the features on the design layers become smaller and approach the resolution of the fabrication process, the consistency between the mask and the printed features layer is significantly reduced. Specifically, it is observed that differences in the pattern of printed features from the mask depend upon the size and shape of the features on the mask and the proximity of the features to one another on the mask. Such differences are called proximity effects.
Some causes of proximity effects are optical proximity effects, such as diffraction of light through the apertures of the optical systems and the patterns of circuits that resemble optical gratings. Optical proximity effects also include underexposure of concave corners (inside corners with interior angles greater than 180 degrees) and overexposure of convex corners (outside corners with interior angles less than 180 degrees), where the polygon represents opaque regions, and different exposures of small features compared to large features projected from the same mask. Other causes of proximity effects are non-optical proximity effects, such as sensitivity of feature size and shape to angle of attack from etching plasmas or deposition by sputtering during the material altering processes, which cause features to have shapes and sizes that have decayed from or accumulated onto their designed shapes and sizes.
In attempts to compensate for proximity effects, the mask layouts can be modified. To illustrate,
In rule-based proximity corrections, corrections such as the serifs and biases of a predetermined size are automatically placed at corners and edges of fabrication layout shapes like the mask shapes 170. Other rules may include adding assist shapes to a mask near desired features from the design layout and adding hammerheads to short end lines of the desired features. Experience of engineers accumulated through trial and error can be expressed as rules, and applied during the fabrication design process. For example, a rule may be expressed as follows: if a feature is isolated, then widen the opaque region in the mask by a particular amount to compensate for expected overexposure, so that the feature prints properly.
The experience captured in rule-based corrections is garnered by going through the fabrication process repeatedly with different mask layouts, making adjustments and observing the results. However, this process consumes time and manufacturing capacity. Even if such experimental, rule-forming runs are made, the resulting rules often do not give satisfactory results as the features become more complicated, or become smaller, or interact in more confined areas, or involve any combination of these effects.
Because rule based corrections are often imprecise and time consuming to revise and test, a proximity effects model is often developed and used to predict the effects of a change in the mask layout with more precision and with fewer off line fabrication runs.
A proximity effects model is typically built for a particular suite of equipment and equipment settings assembled to perform the fabrication process. The model is often built by performing the fabrication process one or a few times with test patterns on one or more mask layouts, observing the actual features printed (for example with a scanning electron micrograph), and fitting a set of equations or matrix transformations that most nearly reproduce the locations of edges of features actually printed as output when the test pattern is provided as input. The output of the proximity effects model is often expressed as an optical intensity. Other proximity effects models provide a calibrated output that includes a variable threshold as well as an optical intensity. Some proximity effects models provide a calibrated output which indicates a printed edge location relative to a particular optical intensity as a spatial deviation. Thus some model outputs can take on values that vary practically continuously between some minimum value and some maximum value.
Once a proximity effects model is produced, it can be used in the fabrication design process. For example, the proximity effects model is run with a proposed mask layout to produce a predicted printed features layer. The predicted printed features layer is compared to the design layer and the differences are analyzed. Based on the differences, modifications to the proposed mask layout are made, such as by adding or removing serifs 193, or adding or removing anti-serifs 197. After making these corrections to the proposed mask layout, a final mask layout is generated that is used in the fabrication process.
A proximity effects model typically produces predictions that lead to corrections that are more accurate than rule based corrections. However, a proximity effects model consumes computational resources by an amount that is related to the number of points of interest where amplitudes are computed. For typical mask layouts, the number of points where the model could be run is large, and the computation time is prohibitive. Therefore it is typical to run the proximity effects model at selected evaluation points located on the edges of features, to determine the correction needed, if any, at each evaluation point, and then to apply that correction to all the points on a segment of the edge in the vicinity of the evaluation point.
If too many evaluation points are selected, the model runs too slowly and the correction procedure takes too long. If too few evaluation points are selected, the model may not detect critical locations where corrections are necessary. Another undesirable effect of too few evaluation points is that segments can become too large, so that even if a correction is accurate for the evaluation point, the correction is applied to too much of the edge. This causes undesirable changes on the edge away from the evaluation point.
What are needed are new ways to select evaluation points and define their vicinity for the mask layouts, so that a reasonable number of comparisons can be made and so that effective modifications to the mask layouts can be suggested.
According to techniques of the present invention, edges of polygons in a mask layout are dissected into segments defined by dissection points, with each segment having an evaluation point, according to modified dissection rules. These techniques allow the spacing of evaluation points and dissection points to be automatically adapted to portions of each edge where changes in the proposed mask layout are most likely needed. According to these techniques, dissection points are closer together where proximity effects are more significant and are farther apart where proximity effects are less significant. Thus unnecessary evaluation points are eliminated and the analysis process is speeded up, while still retaining needed evaluation points.
In one aspect of these techniques, all dissection points can be derived from just a few easily defined parameters. The first parameter is a corner dissection segment length, Lcor, used as a target length to compute dissection points for corner segments that include at least one vertex of the polygon when the length of the edge allows. The second parameter is a detail dissection segment length, Ldet, used as a target length to derive dissection points for segments along portions of an edge where changes in proximity effects are expected to be significant, such as adjacent to corner segments and around projection points when the length of the edge allows. A projection point is a location on an edge where a vertex of another edge comes close enough to introduce sharp changes in the proximity effects. A halo distance is a parameter that determines whether a vertex is close enough to introduce an important effect. A maximum dissection segment length, Lmax, by definition larger than either Lcor or Ldet, is used as a target length to further dissect long portions of the edge that would otherwise not be dissected, such as away from corners and projections points. Extra precision is achieved if a true-gate extension length parameter Lext is used on edges of shifters forming true-gates. The dissection parameters Lcor, Ldet, Lmax, halo and Lext are pre-set and accessed as needed.
In one aspect of the disclosed techniques, a profile of amplitudes from a proximity effects model along an edge of a feature in a design layout is used to dissect the edge. This amplitude profile reflects the level of proximity effects for the given layout and fabrication process. By using this information, a better dissection of the edge is achieved in which the segment length between successive dissection points depends on the magnitude of the proximity effects. For example, on a portion of an edge where the magnitude of the proximity effects is larger than another portion, the segment length is made longer. This kind of dissection avoids making the segment length too short. A segment that is too short, for example in a corner, may result in a very large correction, and may prevent the overall corrections for the layout from converging.
In another aspect, the dissection parameters are derived from the segment lengths based on profiles of amplitudes. According to the modified dissection rules, each evaluation point is placed along a segment at a predetermined fraction of the distance from one dissection point to the other defining each segment, where the predetermined fraction depends on the type of segment. For example, an evaluation point is placed closer to the non-vertex dissection point in a corner segment; and is placed midway between dissection points in other types of segments. Also according to the modified dissection rules, segments are located on polygon edges according to the type of edge, such as whether the edge is a line end, a turn end, or a longer edge. These edge types are deduced from the length of the edge compared to the dissection parameters and whether one or both vertices of the edge are concave or convex corners.
Specifically, according to some aspects of the invention, techniques for correcting proximity effects in a proposed layout corresponding to a design layout for a printed features layer include selecting from among all edges of all polygons in the proposed layout a subset of edges for which proximity corrections are desirable. The subset of edges includes less than all the edges. Evaluation points are established only for the subset of edges. Corrections are determined for at least portions of the subset of edges based on an analysis performed at the evaluation points.
According to other aspects, techniques for correcting proximity effects associated with an edge corresponding to a design layout include establishing a projection point on a first edge corresponding to the design layout based on whether a vertex of a second edge corresponding to the design layout is within a halo distance. The halo distance is associated with significant proximity effects on one edge from a vertex of another edge. The first edge is divided into segments based on the projection point and characteristics of the first edge. It is then determined how to correct the first edge for proximity effects based on the segments.
According to other aspects, techniques for correcting proximity effects associated with an edge corresponding to a design layout include establishing a projection point on a first edge. An evaluation point is determined for the first edge based on the projection point and characteristics of the first edge. It is then determined how to correct at least a portion of the edge for proximity effects based on an analysis at the evaluation point.
According to one embodiment of this aspect of the invention, if the vertex of the second edge is within the halo distance of the first edge, then establishing a projection point includes placing the projection point on the first edge closest to the vertex.
According to another embodiment of this aspect of the invention, determining the evaluation point includes establishing on the first edge two evaluation points spaced apart at least by a target length for detail segments. The two evaluation points straddle the projection point.
According to another embodiment of this aspect of the invention, before establishing a projection point, the halo distance is determined. This includes executing a proximity effects model at several points along a test edge to produce a profile of model amplitudes. A variation in amplitude is determined that is associated with a test vertex on a different test edge. A distance between the test edge and the test vertex is associated with the variation in amplitude. The proximity model is run, the variation is determined, and the distance is associated with the variation for several test edges. Then the halo distance is set based on a distribution of distances associated with variations for the test edges.
According to another embodiment of this aspect of the invention, a computer system for correcting proximity effects associated with an edge corresponding to a design layout includes a computer readable medium carrying data representing the edge, the halo distance, and at least a portion of the design layout. The design layout corresponds to a portion of an integrated circuit. One or more processors are coupled to the computer readable medium. The processors are configured to perform the following steps. A projection point is established on a first edge corresponding to the design layout based on whether a vertex of a second edge is within the halo distance. An evaluation point for the first edge is determined based on the projection point and characteristics of the first edge. A proximity effects model is executed for the evaluation point to produce a model amplitude. A correction is determined for at least a portion of the first edge including the evaluation point based on an analysis of the model amplitude. The correction is then applied to that portion of the first edge.
According to another embodiment of this aspect of the invention, a device is fabricating using a mask for fabricating a printed features layer that includes an opaque region having a segment corrected for proximity effects. The segment corresponds to at least one portion of a first edge in a design layer for the printed features layer. The segment is displaced from the corresponding portion in the design layer by a correction distance. The correction distance is based on analysis of an amplitude output by a proximity effects model at an evaluation point on the corresponding portion. The evaluation point is based on a projection point and characteristics of the edge. The projection point is established based on whether a vertex of a second edge corresponding to the design layout is within the halo distance of the first edge.
The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar elements and in which:
Techniques are described for correcting fabrication layouts for proximity effects during the fabrication of printed features layers, such as in integrated circuits.
The conventional processes include the Functional EDA process 210 that produces the schematic diagram 215. The physical EDA process 220 converts the schematic diagram to a design layout made up of one or more design layers 225 (e.g. design layers 225 a, 225 b, and 225 c). After mask layouts 235 (e.g. mask layouts 235 a, 235 b, and 235 c) are produced, the conventional processes employ a fabrication process 240 to produce the printed features layer 249. The printed features layer 249 may be a layer in a printed circuit or the mask used to produce the layer in the printed circuit. In the former case, the fabrication process includes one process 243 for forming the mask and a second process 245 to produce the layer of the integrated circuit using the mask. If the printed features layer 249 is the mask, step 245 is skipped.
According to techniques of the present invention, the fabrication design process 230 includes a new process 260 for obtaining dissection and evaluation points. In one embodiment, these new dissection and evaluation points are determined for a proposed mask layout 231 and used with a proximity effects model in step 232 to produce a model output 233. The model output is related to the predicted location of edges in the printed features layer if a printed features layer were fabricated with a mask having the proposed mask layout. Hence, in
Each evaluation point is established for a segment of an edge from the proposed mask layout. The segment is defined by a pair of dissection points associated with the evaluation point. Once the dissection points and evaluation points are selected, the proximity effects model is run at the evaluation points. In step 234, the predicted printed features edge 233 for the evaluation point is compared to a target position for that edge in the design layer. Based on an analysis of the comparison, a correction distance is computed. The correction distance is then applied to the entire segment between dissection points associated with the evaluation point on the edge of the affected polygon. As each correction is applied, the mask layout is adjusted in step 236. As a result of the adjustments in step 236 the polygons on the adjusted mask layout may differ from the polygons in the proposed mask layout, as well as from the polygons in the design layers 225. The final mask layouts 235 are then based on the adjusted mask layouts produced by the adjustment process 236.
An overview of one embodiment for the new dissection/evaluation points selection process 260 is illustrated in
For example, in step 390, it is determined whether the current edge is the last edge on the current polygon. If not, process flow passes to step 392 where the next edge requiring dissection on the polygon is selected and made the current edge. Flow then returns to step 330 in which the edge newly made current is dissected according to the dissection parameters. If the edge just processed by steps 330 and 370 is the last edge of the current polygon, process flow passes to step 394 in which it is determined whether the current polygon is the last one needing dissection in the fabrication layout. If not, the next polygon is made the current polygon in step 396, and control passes to step 392 in which the first edge needing dissection on the polygon is made the current edge. If the current polygon is the last on the layout then flow passes to step 232 of
In an alternative embodiment, step 394 includes making another fabrication layout the current fabrication layout and then passing control to step 396. Such a subsequent layout may be necessary to dissect; for example, when the two fabrication layouts are used to doubly expose the same printed features layer, so that edges from the second fabrication layer become part of the printed features layer. This circumstance will be described in more detail later below.
According to one embodiment, the proximity effects model is used to set the dissection parameters, for example, in step 310. These parameters remain constant during the dissection of several polygons in step 330, on one or more proposed fabrication layouts during the fabrication layout design process, step 230.
Herein the term amplitude is used to describe the proximity effects model output associated with a location no matter what model is used. In general, the models are built to accept input representing polygons of opaque or transparent material. In general, an amplitude value is then produced for each location of interest. A threshold amplitude between the minimum amplitude value and the maximum amplitude value can be associated with the model. As used here, the threshold amplitude is associated with the location where the edge of a feature is actually printed. For example, where spatial deviation is output by the model, a threshold of zero marks the location of an edge. As another example, where the model output is simple optical intensity, a non-zero positive value associated with the printed edge serves as the threshold. In some models, the threshold value associated with an edge may itself be a function of several variables such as the optical intensity and intensity gradient at a location.
To understand how dissection parameters can be derived from proximity effects model output, consider
For simplicity of explanation, the proximity effects model that outputs contours like those shown in
The threshold value 470 of about 0.3 model amplitude units corresponds to the value of contour 450 c. Where the model output 465 is above the threshold value 470, a feature is predicted to be printed in the printed features layer; where the model output 465 is below the threshold value 470, no feature is predicted to be printed. Where the model output 465 equals the threshold value is where the edge of the printed feature is predicted to lie. Of course model output units are arbitrary and could be inversely related to what is printed, so that low amplitudes represent printed features and high amplitudes represent no printing. In such a case, the threshold would mark the value below which a feature is printed.
The selection of dissection points and the derivation of dissection parameters from a profile of model output is illustrated in
A profile of model amplitudes along one of these edges reflects the size of proximity effects on that edge for this particular layout and model (where the model represents this particular suite of fabrication equipment and settings). By using information in this profile of amplitudes, a better dissection of this edge can be performed in which the segment length depends on the proximity effect magnitude.
This profile 540 is used to illustrate how particular dissection points are selected on this edge, as well as how the dissection parameters used for dissecting other edges are derived from model output on this edge, according to this embodiment. In other embodiments, variations on the techniques described here are used.
This profile 540 illustrates a length scale of proximity effects at corners of a polygon. That corner length scale is found from a vertex of the edge to a position where the profile flattens near amplitude 0.3. That length scale is about 200 nm. In an alternative embodiment, the corner length scale is selected to be the distance to the threshold value of 0.23, which is about 150 nm. In still another embodiment, the corner length scale is taken to be a distance to an amplitude that is a given percentage of the threshold value. In this case, the given percentage is selected based on how aggressive a correction is sought in the vicinity of corners. Where more aggressive correction is desired, the percentage is chosen to be less than 100%. Where less aggressive correction is desired, the percentage is chosen to be greater than 100%. In still other embodiments, the corner length scale is associated with types of corners, e.g., concave or convex, or angle at the corner, or both. The corner length scale so derived provides a target length to use when dissecting corners where the edge allows.
This profile also suggests a length scale for details near projection points. Considering both sides of each projection point, that detail length scale should be about half the distance over which curve 540 deviates from curves 543 and 544. That detail length scale is about 100 nm. The detail length scale so derived provides a target length to use when dissecting details where the edge allows. This profile also suggests a maximum length scale where the curve 540 hardly varies at all, between the two projection points, of about 300 nm. The maximum length scale so derived provides a target length to use when dissecting edges away from corners and details, where the edge allows.
Along other edges, the distances for the amplitude curve to flatten, the distances over which projection points have noticeable effects, and the length of the undisturbed sections are expected to vary. Thus, dissection parameters based on these must be derived from the observed distributions of length scales. In a preferred embodiment, instead of using a single corner scale from one polygon of one test layout, a statistical corner scale, such as an average corner scale, or median corner scale, is used as a dissection parameter near corners. The statistical corner scale is derived from the corner segment lengths of several polygons on the same test layout or on several layouts. In other embodiments, different corner scales are derived for different corner types, e.g., concave or convex, or different ranges of angles or both. For example, corner segments associated with acute angles less than 45 degrees can be averaged separately from corner segments associated with intermediate angles from 45 through 80 degrees which themselves are averaged separately from corner segments associated with nearly right angles from 80 to 100 degrees.
In another model where proximity effects are different, the values of the length scales and their distributions are also expected to be different. For example, in an equipment suite with lower resolution optics and material processes, the length parameters are expected to be larger. That is, a distance from a vertex of an edge to a flattening point is expected to be greater for all edges tested, and, therefore, the derived dissection parameter is expected to be larger.
In one embodiment, an evaluation point for a corner segment is determined on this curve. According to this embodiment, the evaluation point is placed where the curve has a value equal to a given percentage of the amplitude value where the curve first flattens. For example, if the given percentage is set at 75%, then because curve 540 flattens at about 0.31 amplitude units, the evaluation point is selected where the curve has a value of 75% of 0.31, i.e., a value of 0.23. This occurs at a distance of about 100 nm. In another embodiment, the evaluation point is placed where the profile has a value equal to a given percentage of the threshold value, e.g., 90% of the threshold value of 0.23, which is 0.21. This occurs at about 90 nm on curve 543.
The following paragraphs define particular embodiments for determining dissection points and length scales for dissection parameters. Other embodiments rely on variations of these techniques known in the art.
The 200 nm length of this corner segment is then submitted to a process to derive a corner dissection segment length parameter, Lcor. That process performs a weighted average of this corner length for this edge with other corner lengths from this and other edges to derive the corner dissection segment length, Lcor. In one embodiment, the weights are all the same.
Similarly, moving along edge 511 from its vertex in common with edge 512, at a distance of 1200 nm, the slope first increases above −0.001 at about 1000 nm. That is, the absolute value of the first derivative decreases below the particular value of 0.001 at about 1000 nm. This indicates a corner segment for this edge lies from distance 1000 nm to 1200 nm, and dissection points are placed at both those locations. As above, the 200 nm corner segment length is submitted to the process to derive the corner dissection segment length dissection parameter, Lcor.
According to this embodiment, dissection points are added at each location where the absolute value of slope reaches a local maximum above the small particular value. One embodiment adds dissection points at 553, 551 and 554, which are located at about 300 nm, 480 nm and 860 nm, respectively. Thus, segments that extend from 200 to 300 nm, 300 to 480 nm, and 860 to 1000 nm are added to the two corner segments already defined for this particular edge 511. A residual edge that undergoes further dissection remains between points 551 and 554, from about 480 to 860 nm. Evaluation points are then placed on this edge, one evaluation point on each segment. In this embodiment, each evaluation point is placed midway between the dissection points on each non-corner segment.
The non-corner segment lengths obtained from these embodiments are sent to a process to derive other dissection parameters, such as a detail dissection segment length, Ldet, and a maximum dissection segment length, Lmax. In one embodiment of the process to derive dissection parameters, a histogram of segment lengths is formed and the detail dissection segment length, Ldet, is derived from a mode of the distribution associated with the smallest lengths. In this embodiment, a maximum dissection segment length is derived from a second mode associated with the next smallest lengths. In other embodiments, length scales may be derived from such a distribution of lengths using any methods known in the art. Then, dissection parameters are set based on the derived length scales.
In another embodiment of selecting dissection points from model output along a particular edge, dissection points are also added between these local slope maxima, either at a midpoint, or where the slopes are interpolated to closest approach zero. This embodiment will tend to produce twice the number of segments and cause the dissection parameters derived from the resulting segments to have about half the size. Evaluation points are then placed at the midpoints of the resulting segments.
In still another embodiment, dissection points are added at non-adjacent points where the slope curve 550 has a slope whose absolute value equals the particular minimum slope value, e.g., 0.001. This embodiment adds two dissection points for each local maximum such as at 553 and 554, one on either side of each local maximum.
In another embodiment, before adding dissection points, the second derivative of the profile 540 is computed and examined.
Both projection points are located near points where curve 560 crosses zero. In this embodiment, dissection points are added wherever the second derivative is interpolated to cross zero. This method can be used even for corner segments if the crossing of zero closest to the corner is ignored. Using this method, one corner segment extends from 0 nm to the point indicated by arrow 565, at about 220 nm, and a second corner extends from the point at arrow 568, at about 990 nm to 1200 nm. These lengths of 220 nm and 210 nm, respectively, are sent to a process to derive a corner dissection segment length, Lcor.
This technique also produces dissection points as indicated by arrows 563 at about 320 nm, 561 (associated with projection point 531) at about 480 nm, 566 at about 760 nm, and 564 at about 880 nm. These points yield non-corner segments of lengths 100 nm, 160 nm, 280 nm, 120 nm, and 110 nm. These lengths are sent to the process for deriving non-corner dissection parameters, and support a detail dissection segment length Ldet of about 110 nm, and a maximum dissection segment length Lmax of about 300 nm.
According to one embodiment, an evaluation point is placed at the point on each segment where the absolute value of the second derivative is a maximum. For example, evaluation points are placed at 100 nm, 300 nm, 400 nm, 800 nm, 950 nm, and 900 nm, where the second derivative curve 560 reaches local extreme values between locations where the curve 560 crosses zero.
Using these example embodiments, a proximity effects model is run for a variety of locations along one or more edges of one or more polygons. The polygons can be selected from a proposed or final mask layout, or from a design layer of a design layout, or from a test mask used to build the proximity effects model. The profile of model output along the edge is used to select dissection points and evaluation points for that particular edge. The resulting segment lengths for the particular edge are used with segment lengths obtained for other edges to derive target lengths to use as dissection parameters.
This technique allows the proximity effects model to be run at a few locations, i.e., along certain edges, rather that at every point in a layout. Certain important edges are dissected directly, and the remaining edges are dissected using the dissection parameters. The proximity effects model is then run for all other edges only at the evaluation points. Thus a large number of model runs are avoided and the computation proceeds at a desirable rate. At the same time, the evaluation points are reasonably selected to give the necessary detail on all the edges.
The profiles such as 540 can be generated by any method known in the art. In one embodiment, the test patterns used to build the model are used. These test patterns already include a range of features at different scales and model output computed on a relatively dense output grid. Thus with substantially no additional computations than those already made to build the proximity effects model in the first place, model amplitudes along various polygon edges are already available for deriving dissection parameters.
In another embodiment, the corner length scale is derived from the model output at a series of features of varying pitch. Pitch refers to the smallest distance between repeated features in a layer. The use of features of varying pitch to derive a length scale appropriate for a given model is shown in
In one embodiment, the particular pitch length scale associated with maximum model output at the threshold value is used to set a corner dissection segment length. In another embodiment, this particular pitch length scale is used to set a detail dissection segment length.
In step 710, a profile of proximity effects model amplitudes is generated along at least one edge of a polygon on a test layout. In the preferred embodiment, the test pattern is one used to generate the proximity effects model. In another embodiment, the polygon can be selected on-the fly from a proposed fabrication layout for an actual printed features layer.
In step 730, a corner dissection segment length parameter, Lcor, is set to a distance scale representing the distance from a corner, i.e., a vertex of the polygon defining the current edge, to a point on the profile where the profile flattens. The profile is flat, for example, where successive points undergo less than a certain change in amplitude or absolute value of the slope is less than a particular minimum value. In other embodiments, Lcor is set to a distance scale representing the distance from the corner to a particular percentage of the flat portion, or a particular portion of the threshold, or where the second derivative crosses zero. In some embodiments, a set of Lcor parameters are defined, one for each corner type or angle range or both, and set to appropriate distance scales.
In step 750, dissection points are selected away from the corners, defining non-corner segments, based on changes in model output, such as based on the first derivative or second derivative of the model amplitude profile. This step can also be used to define a halo distance, i.e., a separation between features beyond which no dissection points are necessary on an edge. With a halo distance so defined, if some vertex of another edge is within the halo distance of a non-corner segment of the edge whose profile is being examined, a dissection point should be necessary.
In step 770, the detail dissection segment length parameter, Ldet, is based on the segment lengths of the non-corner segments. For example, the detail dissection segment length is set equal to a mode in the distribution of non-corner segment lengths associated with the shortest segment lengths. For a second example, the detail dissection segment length is set to a particular percentile of non-corner segment lengths, such as the 5th percentile length or 10th percentile length.
In step 790, the maximum dissection segment length parameter, Lmax, is based on the segment lengths of the non-corner segments that are substantially longer than the detail dissection segment length. For example, the maximum dissection segment length is set equal to a second mode in the distribution of non-corner segment lengths, a mode not associated with the shortest segment lengths. For a second example, the detail dissection segment length is set to a particular percentile of non-corner segment lengths, such as the 25th percentile length, the 50th percentile length (the median segment length), or the 75th percentile length.
When the halo and dissection lengths are defined, then control passes to step 330 to use these parameters to dissect polygons in an actual fabrication layout.
In one embodiment, dissection points are based on the vertices of the polygon and five prescribed dissection parameters. The five prescribed dissection parameters include:
corner segment dissection length (Lcor);
detail segment dissection length (Ldet);
maximum segment dissection length (Lmax);
true-gate extension length (Lext); and
halo distance (halo).
The vertices and dissection parameters are first used to determine an edge type. Edges shorter than the shortest prescribed segment length are not worth evaluating and so have no dissection points or evaluation points selected. Edges long compared to twice the corner segment length have a dissection point placed about Lcor from each vertex. A projection point is located on an edge at the closest point to a vertex of another edge, if that vertex is within the halo distance. Long portions of edges away from the corners and without projection points, i.e., residual edges, may be dissected to prevent segments longer than Lmax. Next to corner segments and near projection points, segments should be about Ldet long. Edges of true-gates have lengths increased by the true-gate extension length, Lext, to prevent underexposure of these important features.
In other embodiments, multiple versions of these parameters are defined. For example several Lcor parameters are defined to distinguished convex from concave corners and to distinguish square corners from acute and middle angle corners. As another example, several Lext parameters are defined to distinguish extensions toward an end cap from extensions toward a connector, explained below.
One use of prescribed dissection parameters Lcor and Ldet is illustrated in
There is one evaluation point 840 between each pair of dissection points on an edge. Away from the corner segments, the evaluation points are placed substantially halfway between the dissection points. For example, evaluation point 840 b is halfway between its associated dissection points 830 b and 830 c. This is done because the evaluation point represents the whole segment between the dissection points; and any correction computed for the evaluation point, such as moving the edge x distance units outward, will be applied to the whole segment. Treating the whole segment together is reasonable because the corrections needed are expected to vary slowly along a single segment. That is how the segment length parameters were chosen—to represent a distance associated with each substantial change in proximity effects.
In the corner segments of this embodiment, however, the evaluation points are not halfway. Instead, in this embodiment, the evaluation point is placed two-thirds (⅔) of the way from the convex vertex to the next dissection point. In other embodiments, the evaluation point is placed substantially more than halfway from the vertex to the next dissection point. The placement of evaluation points farther from the vertex is reasonable because the proximity effects in the corner tend to increase more rapidly as the corner is approached. Halfway from the corner the changes are still large compared to the inside dissection point which abuts the next segment. A smaller correction is desired than at the halfway point to avoid big differences with the abutting interior segment.
Effective line-end segment length (dmin);
Effective turn-end corner segment length (dcor); and
Effective interior, flat segment length (dmax).
Line-end segments occur on polygon edges that are no shorter than the shortest prescribed segment length but not long enough to generate more than one segment. Line-ends are likely underexposed or overexposed along their entire length. Such edges are not dissected further but do have an evaluation point. Thus on such edges, the dissection points are the vertices and the evaluation point is placed midway between. The segment and the edge are coincident. Though controlled by the prescribed dissection parameters, the actual segment lengths, dmin, are derived from the vertex positions.
Turn-end edges are also longer than the shortest prescribed length but involve one concave vertex and one convex vertex, such as shown in
Corner, line-end and turn-end dissection lengths are used in those circumstances even if projection points are also present. Projection points control segment lengths in the interior portions of an edge away from the vertices, as will be described later below. Away from corner segments (including all of short edges like line-ends and turn-ends) and projection points, residual portions of an edge are divided evenly so that no segment is greater than Lmax and no segment is shorter than the shorter of Lcor and Ldet. Though controlled by the prescribed dissection parameters Lcor, Ldet and Lmax, the actual length of these segments (dmax)(see dmax 829, for example, in
Line-end segments occur on polygon edges that are no shorter than the shortest prescribed segment length but not long enough to generate more than one segment. Line-ends are likely underexposed or overexposed along their entire length. Such edges are not dissected further but do have an evaluation point. Thus on such edges, the dissection points are the vertices and the evaluation point is placed midway between. The segment and the edge are coincident. Though controlled by the prescribed dissection parameters, the actual segment lengths, dmin, are derived from the vertex positions.
Dissection due to a projection point is illustrated in
In the preferred embodiment, dissection in the vicinity of the projection point is carried out by placing a dissection point 830 q at the projection point 850 and at two other points 830 r and 830 s spaced Ldet 822 from the projection point 850. Evaluation points 840 m and 840 n are then placed at the midpoint of the segments so defined. In another embodiment, an evaluation point is placed at the projection point, and two dissection points spaced Ldet from each other straddle the evaluation point. Such an embodiment is not preferred because it does not allow for different corrections on either side of the projection point as needed for most circumstances, for example, as needed for curve 540 in
Where more than one projection point are initially found on an edge within a distance of Ldet along the edge, only one projection point is kept and used for dissection. Any technique that provides a single projection point from the multiple initial projection points can be used. According to this embodiment, the projection point kept is the one having the shortest distance to the vertex of another edge. After this one is selected, all other projection points on the edge within Ldet of the kept projection point are discarded.
In step 910 of
In step 914 it is determined whether the edge length L, which is necessarily greater than or equal to the minimum dissection length, is also less than double the minimum dissection length. If so, then the edge is a line-end or turn end that is treated as a single segment. Dissection points are selected at the vertices in step 915. Control then control passes to step 370 to select an evaluation point, e.g., in the middle of the edge.
In step 916, the edge length L is necessarily greater than double the minimum dissection length, and it is determined whether the edge is long enough to accommodate two corner segments of length Lcor and an intervening segment of length Ldet. If not, the edge is treated as a line end or as a turn-end. In step 918 it is determined whether the edge is a turn end, i.e., includes both a convex corner and a concave vertex. If not, the edge is a line end and dissection points are placed at the two vertices in step 915 and control passes to step 370 to place the evaluation point. If one vertex is a convex vertex and the other is a concave vertex, so that the edge is a turn-end, it is determined in step 920 whether the edge length L is less than double Ldet. If not, then the edge is split evenly into two segments with dissection points at both vertices and at the mid-point of the edge in step 919. If the edge length L is less than double Ldet, then the edge is kept as one segment with dissection points at the vertices in step 915. In either case, control then passes to step 370 to select evaluation points.
If it is determined in step 916 that the edge length is long enough, two corner segments are placed on the edge, by selecting both vertices of the polygon as dissection points as well as selecting the two points spaced away from the two vertices by the distance Lcor in step 922. A residual edge length is computed by subtracting double the corner dissection segment length, i.e., subtracting 2*Lcor, from the edge length L.
In step 930, one or more projection points are selected on the residual edge, for example, as shown in
In step 940 of
In step 942 it is determined whether the residual length LR is less than twice Ldet. If so, the segment is not further dissected in step 943, and control passes to step 950 to examine the next such pair of dissection points, unless it is determined in step 951 that the current pair is the last pair and control passes to step 370 in
In step 944 the residual length LR is necessarily greater than or equal to double Ldet, and it is determined whether the residual length LR is also less than triple Ldet. If not, the segment is split evenly into two segments by adding a dissection point in the midpoint of the segment in step 945. If so, control passes to step 946.
In step 946 the residual length is necessarily greater than or equal to triple Ldet, and two dissection points are added, each Ldet from one of the current pair of dissection points being examined. Then a new residual edge length is computed for the remaining segment between the two new dissection points. Control then passes to step 948.
In step 948, the new residual edge length is divided by Lmax, and the ceiling integer N of the quotient is obtained. The effective length of the remaining segments, dmax, is computed as the residual length divided by the ceiling integer N. By dividing the residual length by the ceiling integer, dmax is guaranteed to be less than or equal to Lmax. Then N−1 evenly spaced dissection points, spaced apart by dmax, are selected in the remaining segment. Control then passes to step 951 to obtain the next such pair of dissection points in 950 unless the current pair is the last pair and control passes to step 370 in
In one embodiment dissection points and evaluation points are also chosen to be at grid points of a predefined input grid. This additional choice provides the advantage that previously computed model runs can be used for analyzing differences after the edges get dissected.
In yet another embodiment, the proximity effects model is run for closely spaced points along certain critical edges on a proposed layout or layouts, as described above with respect to
This embodiment offers the advantage of customizing the dissection for certain important edges and responding to the actual variations in proximity effects observed along the particular edge.
However, this embodiment requires more computational resources than using dissection parameters because the model is run at many new points along an edge in the fabrication layout for the actual design. This embodiment does not use the pre-computed model values created when the model was built.
Some computation time and power is saved compared to running evaluations at all points on these edges, because a correction is only computed at the more widely spaced evaluation points, one per segment. Once computed, the same correction is then applied to the whole segment. Additional computational time and power can be saved by performing the process only on certain edges expected to suffer from proximity effects, like those on critical features in the layout with dimensions close to the resolution of the fabrication process, or where critical dimensions are within a halo distance of another polygon. For example, this process is used only on edges that include an edge of a true-gate or an edge of a true-gate interconnect.
Some features on a printed layer are formed by exposing the same layer with masks formed according to two fabrication layouts. For example, in a first exposure, narrow phase-shifted features are formed, then, in a second exposure, wider connections between phase-shifted features are formed. Polygons in the first exposure include pairs of phase-shift windows called shifters, which pass light of the same wavelength but out of phase by 180 degrees (or π radians).
Some edges have portions that are printed and portions that are not printed. For example, no edge is printed between points 1043 and 1042 of the polygon 1020 edge connecting points 1043 to 1040. Similarly, no edge is printed between points 1038 and 1042 of shifter 1010 a edge connecting point 1038 to point 1031. Such partially printed edges usually involve a point that marks the intersection of a shifter edge with a trim edge, like point 1042.
According to these embodiments, dissection points and evaluation points are not placed on edges that are not printed, such as outside edges polygons for shifters or edges of trim polygons inside shifters. For example, when dissecting the polygon of shifter 1010 a, the edge connecting vertices 1031 and 1036 contains portions that will be printed, therefore this edge is dissected. However, the edge connecting vertices 1037 to 1038 will not be printed; and therefore it is not dissected.
Also according to these embodiments, dissection points and evaluation points are moved or removed to ensure that segments do not have lengths less than a minimum segment length, and that the evaluation points are on the physically important parts of the segments. According to one embodiment, dissection points are initially placed at the corners of the true-gate, i.e., at points 1032 and 1035 in addition to shifter vertexes 1031 and 1036. The edge of the true-gate is then dissected as if 1032 and 1035 were vertices of a polygon. For example, normally a dissection point is placed at point 1032 and another dissection point is placed a distance Lcor away at 1033, and the evaluation point 1034 is placed two-thirds of the way from point 1032 to point 1033.
However, if the length of the resulting segment from vertex 1031 to true-gate corner 1032 is less than the minimum segment length, e.g., the lesser of Lcor and Ldet, a dissection point is not placed at true-gate corner 1032. Instead, the dissection point at 1031 is used. The evaluation point 1034 for the segment remains on the true-gate edge two-thirds of teh way between points 1032 and 1033. That is, the evaluation point is located as if the corner of the true-gate 1032 were a corner dissection point. Because the segment from vertex 1031 to point 1032 is used to connect the corner of the shifter to the true-gate, it is referred to as the shifter extension, and the shifter extension length is the distance from shifter vertex 1031 to true gate vertex 1032.
This technique has the advantage of simultaneously keeping all segment lengths greater than or equal to the minimum segment length, but ensuring that the evaluation points are on the true-gate, the critical feature in the layout. This is an advantage because it is desired that the correction determined at the evaluation point be applicable to the true-gate, which is the important feature. The dissection of an entire edge, like shifter polygon edge from vertex 1031 to 1036, considering the influence of true gate corners is implemented during an alternative embodiment of step 330 of
Where a portion of an edge is not printed, dissection points are selected to avoid segments shorter than the minimum dissection segment length and evaluation points are placed on the portion being printed. According to one embodiment, dissection points are initially placed at the intersection of shifter and trim edges in addition to trim edge vertexes. For example, a dissection point is initially placed at intersection point 1042 and at point 1041, a distance Lcor away. The evaluation point is then placed two-thirds of the way from point 1042 to point 1041, as if the intersection point 1042 were a vertex. However, if the non-printed portion of the edge, from point 1043 to 1042, is shorter than the lesser of Lcor and Ldet, the segment from 1041 to 1042 is extended to point 1043. That is, the segment is extended to the trim 1024 a by the trim extension distance from 1042 to 1043: The evaluation point remains unchanged, at two-thirds the distance from the intersection point 1042 to the dissection point 1041. The remaining edge of the polygon 1020 is then dissected according to the rules stated above for non-corner segments.
This technique has the advantage of simultaneously keeping all segment lengths greater than or equal to the minimum segment length, but ensuring that the evaluation points are on the printed portion of the edge. This is an advantage because it is desired that the correction determined at the evaluation point be applicable to the printed edge.
Other arrangements yield other results. For example,
If the distance between true-gate corners 1055 a and 1055 b is greater than or equal to the minimum segment length, then the dissection of this edge of polygon 1021 connecting vertices 1054 f to 1054 g proceeds as follows. Normally true-gate corners 1055 a and 1055 b are dissection points and the evaluation point 1056 is midway between the two, as if the true-gate corners 1055 a and 1055 b were polygon vertices on a line-end. However, if both edge portions from true-gate corner to edge vertex, i.e. edge portion from vertex 1054 f to true-gate corner 1055 a and edge portion from true-gate corner 1055 b to vertex 1054 g, have lengths shorter than the minimum segment length, then the dissection points are moved to the vertices. The evaluation point 1056 remains unchanged, midway between the corners of the true-gate. If only one of these edge portions from true-gate corner to vertex is less than the minimum, the dissection point is moved to the vertex for the short edge portion. For example, if only the edge portion from vertex 1054 f to true-gate corner 1055 a has a length less than the minimum dissection length, then the dissection point is moved from 1055 a to 1054 f, while true-gate corner 1055 b remains a dissection point. The evaluation point is moved so that it is placed two-thirds of the way from true-gate corner 1055 a to 1055 b.
These techniques illustrated with respect to
Other edges of polygon 1021, such as those connecting vertices 1054 b to 1054 c, 1054 c to 1054 d, 1054 d to 1054 e, and 1054 e to 1054 f are inside shifter 1011 a and are not printed. Also edges of shifters outside the trim, such as edges connecting vertices 1053 c to 1053 d, 1053 d to 1053 e, 1053 e to 1053 f, and 1053 f to 1053 a are not printed. Therefore these edges of polygon 1021 and shifter 1011 are not dissected. Still other edges, such as that edge of polygon 1021 connecting vertex 1054 a to 1054 b, are partially printed and are handled as described above with respect to
In order to print a true-gate area more aggressively, to ensure it will not be underexposed, a true-gate is often artificially extended. The amount that a true-gate is extended during design of a fabrication layout, such as a mask layout, can be specified by a designer with a true-gate extension length parameter Lext. In some instances, the designer may specify several true-gate extension length parameters for different circumstances. For example, a designer may specify a first true-gate extension for use where a true-gate is connected to another element and a second true-gate extension for use where a true-gate is capped by an end-cap.
The actual true-gate extension depends both on the parameters Lext and the actual feature geometries. For example, with respect to
According to this embodiment, the dissection points are selected to be consistent with the true-gate extension scheme employed in the fabrication layout. For example, the dissection of an edge with a true gate described above is begun after the true-gate is extended to have corners at points 1031 and 1039.
If Lext parameters have not been defined, then dissection proceeds as follows. Normally, dissection points are placed at the corners of the true-gate. For example, dissections points are placed at 1065 a and 1065 b for true-gate 1003 a, and at 1065 c and 1065 d for true-gate 1003 b. In addition, dissection points are placed a distance Lcor from the corner points of the true-gates. For example, a dissection point is placed at 1063 b, a distance Lcor from true-gate corner 1065 b, for true-gate 1003 a. Similarly, a dissection point is placed at 1063 f, a distance Lcor from true-gate corner 1065 c, for true-gate 1003 b. The evaluation point is placed two-thirds of the way along the resulting segment from the true-gate corner. For example, evaluation point 1066 a is placed two-thirds of the way from true-gate corner 1065 b to dissection point 1063 b. Similarly, evaluation point 1066 b is placed two-thirds of the way from true-gate corner 1065 c to dissection point 1063 f. No evaluation point is placed on the edge portion between the two true-gates, from point 1065 b to point 1065 c, because that edge is not printed and does not require correction.
However, if the distance between the true gates is less than the minimum segment length, e.g., the lesser of Lcor and Ldet, then the dissection points are moved to the middle point of the gap. For example, if the distance from true-gate corners 1065 b of true-gate 1003 a to true-gate corner 1065 c of true-gate 1003 b is less than the minimum of Lcor and Ldet, then the dissection points at 1065 b and 1065 c are both moved to point 1063 d. The evaluation points are not moved. They remain as if the corners of the true-gate were vertices of the polygon.
If a Lext parameter has been defined, then dissection proceeds as follows. True-gate 1003 a is extended to point 1063 c, a distance Lext for end caps from the original true-gate corner at 1065 b. Similarly, true-gate 1003 b is extended to point 1063 e, a distance Lext for end caps from the original true-gate corner at 1065 c. No evaluation point is placed on the edge portion between the two true-gates, from point 1063 c to point 1063 e, because that edge is not printed and does not require correction. Similarly the other corners of the true-gates, 1065 a and 1065 d are moved outward a distance Lext for connectors to points 1067 a and 1067 b, respectively. In some embodiments, Lext for end caps is equal to Lext for connectors; in other embodiments they have different values. The evaluation points are placed two-thirds of the way from the extended corners 1063 c and 1063 e, respectively, to the dissection points 1063 b and 1063 f, respectively, each dissection point spaced a distance Lcor from the extended corners.
However, if the distance between the extended true-gates, from point 1063 c to point 1063 e, is less than the minimum dissection length, then the dissection points are not placed at 1063 c and 1063 e, but are both moved to the center point 1063 d. The evaluation points are not moved.
The techniques described with respect to
In step 1110 it is determined whether all of an edge is important for the edge on which dissection points and evaluation points were most recently selected. If all of the edge just completed is important, then control passes to step 1120 where the next edge on the polygon is made the current edge. If some of the edge is less important, such as because some of the edge is not a true-gate or some is not printed, then flow passes to step 1115 in which additional dissection points are considered, such as at intersections of polygon edges with shifter edges, and at corners of true gates. Evaluation points are only defined on important portions of segments. After these steps are performed, then flow passes to step 1120 to make the next edge the current edge.
In step 1125 it is determined whether any of the next edge is printed. If not, flow does not return to step 330 (of
If at least some of the edge is printed, flow goes eventually to step 330 to select those dissection points on the next edge.
Once the dissection points and associated evaluation points are selected, the model is run for the actual layout only for the evaluation points, as shown by process 232 in
Once the segments have been moved by the correction distance for all the affected polygons, the adjusted fabrication layout (produced by process 236 in
In another embodiment, the adjusted fabrication layout is tested again with the proximity effects model to determine whether the agreement between the printed features layer (249 in
In one embodiment, only the new edges of the polygon generated by moving certain segments by the corresponding correction distances are dissected. In this embodiment, the halo used to find projection points is also reduced for the next round of evaluations, so that fewer corrections are attempted each round, and the solution can converge. This is accomplished, for example, as shown in
Computer system 1200 may be coupled via bus 1202 to a display 1212, such as a cathode ray tube (CRT), for displaying information to a computer user. An input device 1214, including alphanumeric and other keys, is coupled to bus 1202 for communicating information and command selections to processor 1204. Another type of user input device is cursor control 1216, such as a mouse, a trackball, or cursor direction keys for communicating direction information and command selections to processor 1204 and for controlling cursor movement on display 1212. This input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), that allows the device to specify positions in a plane.
The invention is related to the use of computer system 1200 for producing design layouts and fabrication layouts. According to one embodiment of the invention, dissection points and evaluation points for fabrication layouts are provided by computer system 1200 in response to processor 1204 executing, e.g., as threads, one or more sequences of one or more instructions contained in main memory 1206. For example, the evaluation point determining process runs as a thread 1252 on processor 1204 based on evaluation point determining process instructions 1251 stored in main memory 1206. Such instructions may be read into main memory 1206 from another computer-readable medium, such as storage device 1210. Execution of the sequences of instructions contained in main memory 1206 causes processor 1204 to perform the process steps described herein. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions to implement the invention. Thus, embodiments of the invention are not limited to any specific combination of hardware circuitry and software.
The term “computer-readable medium” as used herein refers to any medium that participates in providing instructions to processor 1204 for execution. Such a medium may take many forms, including but not limited to, non-volatile media, volatile media, and transmission media. Non-volatile media includes, for example, optical or magnetic disks, such as storage device 1210. Volatile media includes dynamic memory, such as main memory 1206. Transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise bus 1202. Transmission media can also take the form of acoustic or light waves, such as those generated during radio-wave and infra-red data communications.
Common forms of computer-readable media include, for example, a floppy disk, a flexible disk, hard disk, magnetic tape, or any other magnetic medium, a CD-ROM, any other optical medium, punchcards, papertape, any other physical medium with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave as described hereinafter, or any other medium from which a computer can read.
Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to processor 1204 for execution. For example, the instructions may initially be carried on a magnetic disk of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to computer system 1200 can receive the data on the telephone line and use an infra-red transmitter to convert the data to an infra-red signal. An infra-red detector can receive the data carried in the infra-red signal and appropriate circuitry can place the data on bus 1202. Bus 1202 carries the data to main memory 1206, from which processor 1204 retrieves and executes the instructions. The instructions received by main memory 1206 may optionally be stored on storage device 1210 either before or after execution by processor 1204.
Computer system 1200 also includes a communication interface 1218 coupled to bus 1202. Communication interface 1218 provides a two-way data communication coupling to a network link 1220 that is connected to a local network 1222. For example, communication interface 1218 may be an integrated services digital network (ISDN) card or a modem to provide a data communication connection to a corresponding type of telephone line. As another example, communication interface 1218 may be a local area network (LAN) card to provide a data communication connection to a compatible LAN. Wireless links may also be implemented. In any such implementation, communication interface 1218 sends and receives electrical, electromagnetic or optical signals that carry digital data streams representing various types of information.
Network link 1220 typically provides data communication through one or more networks to other data devices. For example, network link 1220 may provide a connection through local network 1222 to a host computer 1224. Local network 1222 uses electrical, electromagnetic or optical signals that carry digital data streams. The signals through the various networks and the signals on network link 1220 and through communication interface 1218, which carry the digital data to and from computer system 1200, are exemplary forms of carrier waves transporting the information.
Computer system 1200 can send messages and receive data, including program code, through the network(s), network link 1220 and communication interface 1218.
The received code may be executed by processor 1204 as it is received, and/or stored in storage device 1210, or other non-volatile storage for later execution. In this manner, computer system 1200 may obtain application code in the form of a carrier wave.
In the foregoing specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.